MLU/MLU-3 combine interrupted utterances

[nabil...@gmail.com]

Dear all,

I was working on MLU-3 analysis and realized that the programme was unable to run MAXWD to extract the "longest utterance" when "+/." and "+," has been added to interrupted utterances.

Instead, something like this (a partial utterance) would be extracted instead:

*CHI: +, a cat is at [x 3] the beach .

I was wondering if there is a command that I can input to allow for the programme to combine these utterances into a single utterance for MLU analysis?

Looking forward to your responses.

Thank you!

Best Regards,

Nabilah

[Leonid Spektor]

Hi Nabilah,

Sorry for the late reply. I had to consult with other people here about this. Unfortunately, there is no way to do what you want to do with MAXWD at this time. Perhaps you could explain in more detail what is your goal in trying to use MAXWD this way. Maybe that will help people in charge here to appreciate this more.

Leonid~

On Jan 20, 2021, at 02:41, nabil...@gmail.com <nabil...@gmail.com> wrote:

Dear all,

--
You received this message because you are subscribed to the Google Groups "chibolts" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chibolts+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/6405c970-5452-4043-997a-cffb78c9de98n%40googlegroups.com.

[Nabilah Mohamed Zaini]

Dear Leonid,

No worries. Thank you so much for your input.

Within our study, utterances are being split based on communication units that consist of the main clause and its dependent clauses. However, some utterances have been split due to interruptions by the administrator (e.g. prompts, affirmations) have been re-combined using the "+/." and "+," markers between utterances as it forms a communication unit when combined.

*CHI: one sunny sunday morning the cat +/.
*CHI: +, is at [x 3] the beach .

Based on the CHAT manual, it seemed like it was possible to combine these utterances using the MLU function: "An advantage of using +/. instead of +... is that programs like MLU are able to piece together the two segments and treat it as a single utterance when a segment with +/. is followed by +, on the next utterance."

Due to the population that we are testing, we intend to use MLU-3 instead of MLU for our analysis. Following the instructions from CLAN, we would have to run MAXWD on these files first: "The second CLAN analysis we will perform computes the mean length in morphemes of each child’s five longest utterances. To do this, we will run MAXWD on the five files in the ne20 folder and then MLU on the output of MAXWD. By default, MAXWD runs on the %mor line, rather than the main line. maxwd +t*CHI +g1 +c5 +d1 *.cha"

For our analysis, we have used these commands for MLU-3:

1. maxwd +t*CHI +g1 +c3 +d1 +f +s+xxx *.cha

2. mlu +s+xxx *.cex

 However, in the event of split utterances (due to interruptions), this output 3 longest utterance will be extracted instead:

*CHI: cat is trying to catch the butterfly.
*CHI: the [x 3] boy is trying to &-s say next time cannot catch the butterfly +/. (The utterance combined to this, which is the next child utterance within the transcript, has not been combined automatically.)
*CHI: +, is at [x 3] the beach . (The utterance combined to this, which is the previous child utterance within the transcript, has not been combined automatically. Refer to above example for the combined utterance.)

 (Note: Analysis tiers were excluded, this is just an example.)  

As shown, this leaves us with partial utterances which would not be accurate for an MLU-3 analysis.

Hope this clarifies the situation, and I hope that you could advise if there is a way to resolve this situation.

Thank you!

Best Regards,

Nabilah

You received this message because you are subscribed to a topic in the Google Groups "chibolts" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/chibolts/62kSfaPuEXc/unsubscribe.
To unsubscribe from this group and all its topics, send an email to chibolts+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/820AF019-0E11-48E3-97C7-9E66A1316F4A%40andrew.cmu.edu.